Technology, Tools, and Talent: what you need to succeed as a leader in global development

IDinsight CEO Rebecca Sharp was invited to give a keynote address at Georgetown University’s McCourt School of Public Policy as part of its Fall 2024 Master of International Development Policy (MIDP) Program. Here are highlights from her address.

In today’s ever-changing global development landscape, leaders need to master three things to succeed: technology, tools, and talent. By “technology,” I’m referring to the development of data science and artificial intelligence and their applications to the world’s most pressing problems, like poverty and unequal access to information. By “tools,” I mean the rigorous evidence-based tools that allow us to understand what is working to solve poverty and what is not. And by “talent,” I mean the soft skills of how to cultivate workplace teams and cultures that motivate, inspire, and push smart people to solve hard problems.

At IDinsight, we have invested a lot in focusing on all 3 of these “Ts.” 

Technology

The technological revolution is happening faster and faster: the agricultural revolution took several millennia, the industrial and mechanical revolutions took a century, the digital revolution took half that, and now the artificial intelligence revolution is happening in a matter of years. Technological progress has never been faster than it is now.

What’s exciting is the potential that tech has to democratize valuable information that was previously only available in high-resource settings. What if everyone could someday have an AI teacher’s assistant in their classroom, an AI community health worker to assist with accurate diagnosis and treatment, or an AI agricultural extension worker in their pocket?

These things are not just dreams. They’re actually being built by inspiring social entrepreneurs, including  IDinsight and our partners. For example, IDinsight is partnering with Reach Digital, an NGO in South Africa, to improve maternal health chatbots using Ask A Question, an AI-powered question-answering service built by IDinsight’s Data Science and Engineering team. 

How does it work? Young mothers have a lot of questions about their maternal and child health, and the maternal and infant mortality rates in South Africa remain stubbornly high in many areas. The government set up a hotline for mothers to ask their questions, but it was fielding 1,200 questions a day and only staffed by 3 people. IDinsight built an easy-to-use WhatsApp chatbot that responds to mothers’ health questions, like “Where do I go to get my baby immunized?” or “Can I take this medicine at this stage of my pregnancy?” It escalates roughly 30% of inquiries to a human, who can now dedicate their time to really life-threatening or complex inquiries. The rest of the responses match government-approved databases of medical content, so it does not ‘hallucinate’ (aka makeup answers). 

Another similar Large Language Model (LLM) tool that our Data Science team built is “Ask A Metric.” This is a tool that uses a Gen AI WhatsApp chatbot to query a decisionmaker’s full set of databases. With this, you don’t have to sift through reams of paper or ask somebody to track down the information you need for a meeting in 5 minutes or to make a critical decision about resource allocation: the chatbot will tell you in 30 seconds or less. Check out a video demo here.

Learn more about our technological innovations here.

Tools

“Tools” refers to rigorous data and evidence tools rooted in a counterfactual and attributable view of impact. New technological applications are great, but they need to be tested rigorously to ensure they are actually achieving a positive impact and are cost-effective.

Rigorous measurement is in our DNA at IDinsight – we bring impact measurement tools to decision makers. Over a dozen years ago when we were founded, RCTs were typically only run in academic or knowledge-focused settings. 13 years later, IDinsight has completed many rigorous, decision-focused impact evaluations and contributed to many practitioners’ learning agendas, as well as the scale-up of successful interventions in a variety of sectors. 

Our wise Board member Matt Forti from One Acre Fund likes to say, “Data and evidence tools can be used to prove or to improve,” – at IDinsight, we’ve done both, but we’re most proud of the many examples of where our measurement has allowed clients to change course, and try new ways of working that ultimately increase their impact. At the end of the day, new-age technology and rigorous counterfactual tools are both necessary to deploy at the right times in an organization or intervention’s scale journey. 

Take Educate Girls as an example. Educate Girls is an Indian NGO that is tackling the massive problem of 3 million girls who are out of school and simply tries to get them back in school using a multifaceted approach, including community training, parent visits, and life skills training. IDinsight conducted a 3-year RCT for Educate Girls that undergirded one of the world’s first Development Impact Bonds in education. The RCT found that enrollment and learning results exceeded the impact targets the funders had set, and as a result, Educate Girls received a $100M Audacious investment. 

But our partnership didn’t end there. With this cash infusion, Educate Girls has to use that money to execute very ambitious plans to reach millions more children over the next five years. So the next question became, “How do we efficiently find more of these out-of-school girls”? IDinsight developed a machine learning algorithm that we trained on the extensive data Educate Girls had collected about out-of-school girls from their previous programming and then applied that data predictively to new districts. This cost-effective algorithm allowed Educate Girls to find 600,000 more out-of-school girls than what they would have found without our predictions. 

Check out more of our tools here.

Talent

I strongly believe that you can have the tech and the tools, but leaders in global development will really struggle to maximize their impact if they cannot figure out how to build great team cultures, make people feel valued, inspired, and motivated to solve hard problems and rise up in the extremely competitive market for talent. 

Compared to the “hard skills” required for building technology and tools, “soft skills” are somewhat harder to understand and implement because one doesn’t commonly learn them in the classroom. And these “soft skills” are not limited to CEOs or future C-suite executives but are applicable even if you are an individual researcher, no matter what role you occupy on a team; and definitely to middle managers and supervisors.

Here are a few of my top “leadership rules,” all of which boil down to critical soft skills: 

• Create an environment of psychological safety: Google did a large-n research across all of its teams and found that psychological safety was the single highest predictor of great, high-performing teams. Psychological safety ensures great ideas get brought to the forefront instead of being buried. Harvard Business Review has one of my all-time favorite articles on this, called “The problem with saying my door is always open.” It’s about how leaders should not expect others to come to them with their problems – that fundamentally ignores the dynamics of power structures and hierarchy within an organization. Instead, leaders should go to people in the organization to proactively solicit feedback. You can’t be a good listener if no one feels comfortable talking to you.
• Give and receive feedback well: Giving and receiving feedback are two of the most important skills that define effective leaders. On giving feedback, a 5:1 positive to constructive ratio is what evidence shows is necessary for people to feel trust in both personal and professional relationships. Having a foundation of trust makes it easier, not harder, to be transparent and direct when something hard needs to be conveyed. Holding back when something is going wrong is a huge disservice to teams. Learning how to receive feedback is an important skill as well. It can be learned and is almost a formula! One needs to start by not responding with a retort. Instead, listen to what the person is saying, nod, make eye contact, and listen actively. Proceed by asking clarifying questions and then summarizing your understanding of what was said. Only after you’ve done all these things and thanked the person prolifically for their gift of feedback can you cautiously offer them some additional information that might help them understand while again not taking away from the unintentional impact you had on them.
  
• Navigate cultural differences with sensitivity: The Culture Map outlines some social science research around different ways that cultural differences manifest in the workplace, including power distance/hierarchy, perceptions of time, decision-making processes, etc. Many ways that you may think of “success” or “high performance” or even appropriate ways to react in the workplace are very different from what others think! At IDinsight, we’ve grappled with ways that we can adjust our performance rubrics to take into account diverse ways of working, i.e., not requiring high performance be measured by being the most outspoken or self-promoting person (a style that is counterintuitive in some non-Western cultures), but similarly rewarding a wide range of positive traits including humility, deep listening, ability to build strong client relationships, and cultural sensitivity.
• Delegate effectively: I recommend “Managing to Change the World” and The Management Center, which have lots of free and very actionable resources for how to delegate, decision-making modes, check-in guides, and how to manage other tricky leadership and supervisory issues.

Read more about how we are reducing power asymmetries in the social sector.